Coding Errors and Classification Refinement
研究了编码错误如何扭曲费率制定中的分类相对值，提出了评估其重要性的方法，并给出了减少影响的建议。
Data used to determine class relativities for ratemaking purposes are not free from error. Errors in coding classification variables on reports relating to exposures, premiums, or losses are of substantial concern because their impact is not eliminated by the existence of large numbers. This paper presents a model of the way in which coding errors distort the computed class relativities. The question of whether the distortion caused by coding errors is material enough to cause concern is addressed, and a way of assessing the materiality of coding errors is developed. Some methods for reducing the impact of data coding errors on the resulting relativities are suggested.